Should an AI scribe pick my billing code?

Short answer: no. An AI scribe can suggest a billing level or diagnosis code beside the note, with its reasoning shown, but it should never write one into the note for you. A code the software picked is still your code, and the clinicians who have checked these tools say the same thing: review every one.

What can AI scribes do with codes today?

• Suggest a billing level, with a written justification.

• Update a diagnosis and add supporting criteria from the visit.

• Pick diagnosis codes.

• Enter medication orders from the note, sometimes with default doses.

A clinician running a psychiatric practice describes the first two, and loves them:

“You can tell it to update a diagnosis and add the supporting DSM5 criteria from the visit to support it. You can also ask it to suggest the billing complexity based on MDM and add a justification to support the code.”

MDM is medical decision-making. It is easy to see the appeal: those are two of the most tedious paragraphs in psychiatry, and the tool writes both.

How accurate are AI-suggested codes?

Accurate enough to save time, not accurate enough to trust unread. A clinician who sat through a scribe demo found the vendor's own reviewers picking the diagnosis codes:

“With little training in that department so you have to go back and make sure the codes are correct.”

Another, using the scribe built into their records system, is blunter:

“It picks the wrong things, uses the wrong diagnosis codes.”

A pain medicine clinician who has built medical-necessity language into their templates says they still read every note before it is submitted.

Who is responsible for a wrong billing code?

You are. A wrong word in a history costs two seconds. A wrong code costs a denial, a clawback or an audit, and every one arrives with your name on it, not the vendor's. Who answers for the note has one answer, and it is not the software.

What should a good AI scribe do with codes?

That last row is not hypothetical. One clinician whose system enters medication orders from the note warns:

“It'll use default dosages so you need to specify that in your note.”

How do I check AI-suggested codes quickly?

1. Read the note first, then the code. The code should follow from what is written.

2. Check the billing level against the time and complexity you actually documented.

3. Match each diagnosis code to something you said in the visit.

4. Say doses out loud, so the software never has to pick one.

5. Keep the rule the careful clinicians keep: nothing is submitted unread.

Frequently asked questions

Can an AI scribe do my medical billing?

It can draft the pieces: a suggested level, a justification, candidate diagnosis codes. It cannot take responsibility for them. Treat every suggestion as a draft you approve, the same way you treat the note.

Will an AI scribe document medical necessity?

It can include the language if you build it into your template, but a payer reads necessity as an argument, not a transcript. See will an AI scribe document medical necessity.

Does Solanto suggest billing codes?

Solanto will document, not code. It will write the history, leave the judgment to you, and keep anything it suggests out of the note until you put it there.

The short version

Let an AI scribe suggest a code beside the note, with its reasoning. Never let it write one in for you. Solanto will document, not code: it will write the history, leave the judgment to you, and keep anything it suggests out of the note until you put it there.

The price is on the page, not behind a call. Early access is open: solanto.ai/pricing

Ask a colleague whose scribe suggests codes how many they overruled this month. If the answer is none, ask whether they are reading them.